SynLlama represents a significant advancement in drug discovery by fine-tuning Llama3 LLMs to generate practical, synthesizable molecules with complete synthetic pathways.

• Creates molecules using common building blocks and robust reaction templates
• Enables faster drug development by focusing on synthesizable compounds
• Generates diverse molecular analogs with similar properties
• Bridges the gap between AI-generated molecules and practical pharmaceutical applications

This innovation addresses a critical challenge in medicinal chemistry by ensuring AI-generated drug candidates can actually be produced in the lab, potentially accelerating development timelines and reducing costs in pharmaceutical research.

SynLlama: Generating Synthesizable Molecules and Their Analogs with Large Language Models